Google's Gemini 3.5 Pro was promised for June. At the end of September, its API changelog — latest entry dated September 22 — still does not list it.

That delay sits at the center of a new analysis by Pedro Dias, published on Search Engine Journal and republished from The Inference. Dias, who says he has spent 20 years in search and bet on Google for years, has changed his position: "I think Google's chances of leading in AI are slim."

His core argument: Google lacks the motivation to build the product most likely to undermine its search business, because search still works. "The question was always whether Google is willing to break something that prints money," he writes.

A missed date and a coding gap

In February, Gemini 3.1 Pro led the Artificial Analysis Intelligence Index, beating Claude Opus 4.6 by four points at less than half the running cost. At I/O in May, Google promised Gemini 3.5 Pro for June. In mid-July, Bloomberg reported the model's coding fell short and that a training-data fix had disappointed. Google's statement emphasized shipping many models quickly and keeping them "highly cost-effective for customers."

Google has shipped plenty since July: Gemini 3.7 Flash, 3.8 Flash, speech, video, music and robotics models. But in Q2 earnings remarks, Sundar Pichai was already talking about Gemini 4 and "our most ambitious pre-training run yet" — before 3.5 Pro has shipped.

The chips tell the story

The scarce resource in AI is compute, and Dias follows where Google's TPUs go. In May, The Information reported Anthropic committed to spend $200 billion over five years on Google's cloud and chips. In Q2, Google Cloud revenue grew 82% to $24.8 billion, with $8.8 billion in operating income.

Set against that, a Reuters exclusive from August: tight compute capacity and internal disputes left areas such as coding short of resources, with Cloud's senior leaders hoping new Gemini leadership would calm a fight with DeepMind over chip allocation. In April, The Information reported a memo in which Sergey Brin told DeepMind staff Google must "urgently bridge the gap" with Claude on agentic coding.

Dias draws the business logic sharply: "A TPU rented to Anthropic is revenue this quarter. A TPU spent on another Gemini training run is a bet." A company that must win takes the bet, he argues; a company already winning takes the contract.

Consumer gains are real. Enterprise is another matter

Dias concedes Google's strengths. Its cloud backlog stood at $514 billion at the end of Q2. The Gemini app passed a billion monthly users in August, and Pichai says AI Mode has passed a billion too. Comscore's U.S. panel shows Gemini's share of AI assistant prompts rising from 17% in January to 30% in June, while ChatGPT fell from 70% to 50%.

But where paying customers choose, Google trails. Menlo Ventures' 2025 enterprise report put Anthropic at 40% of enterprise LLM API spend, OpenAI at 27% and Google at 21%. In coding, Anthropic held 54%, OpenAI 21%, Google 11%. Dias notes Menlo is one of Anthropic's biggest backers and co-runs a fund with it — "read those numbers with that in mind" — yet concludes the paid-work money has mostly gone elsewhere.

The Microsoft precedent

Dias sees a cautionary parallel. Microsoft had Windows, Edge and Office as default placement and moved first, putting OpenAI's technology into Bing in February 2023. Three and a half years later, StatCounter's August figures show Bing at 4.5% of worldwide search, and Copilot sending 2.8% of tracked AI chatbot referral traffic. Microsoft thrives renting cloud compute to model builders — including OpenAI — but, as Dias puts it, "nobody calls it the leader in search or consumer AI."

The pattern repeats in Google's own history: Google+ launched in 2011 with Gmail, Android and YouTube behind it, and shut its consumer version in 2019. Google researchers published the transformer paper in 2017; OpenAI turned it into ChatGPT.

Talent flows point the same way. Noam Shazeer, co-lead of Gemini, left for OpenAI in June. In August, Jeff Dean departed after 27 years to co-found Discovery Loop, Demis Hassabis stepped back to chair DeepMind, and Koray Kavukcuoglu took direct control of Gemini. Fortune, using Zeki Data, reports DeepMind's hires per departure fell from roughly 12 in 2023 to roughly two now, while Anthropic managed 22 in 2025.

Search still pays

Google's AI defense of Search is working financially. Search and other revenue grew 17% in Q2 to $63.3 billion, and Pichai credits AI features with driving query growth. Dias argues that comfort is precisely the problem: the real threat is an assistant that completes the job without ever showing a results page — and every incentive inside Google points away from building it.

He allows that Google can sprint when frightened, citing the December 2022 code red after ChatGPT and Gemini 2.5 Pro topping LMArena in March 2025. But "nobody declares a code red when Search revenue is up 17%."

Dias names two signals that would change his mind: a Gemini product that does the job end to end with no results page and no ad slot, and Google celebrating that product taking queries away from Search on an earnings call. For marketers watching referral traffic and AI-surface economics, his benchmark is the one to watch.